Three Carnaps on Ontology - Northern Arizona University Three... · 7 There might be no ... rather...

OUP CORRECTED PROOF – FINAL, 12/17/2015, SPi

Carnap is frequently portrayed as advocating the elimination of metaphysics. Here, e.g., is Robin Le Poidevin:

The classic paper that sets out to undermine ontology is Rudolf Carnap’s “Empiricism, Semantics, and Ontology” . . .1

Carnap’s goal, according to this prevalent picture, is to discredit ontology: to encour-age us to stop doing it. Huw Price speaks of “. . . [Carnap’s] celebrated attack on meta-physics;”2 Matti Eklund addresses “the skeptical or deflationary or dismissive attitude toward ontology that Carnap seems to have;”3 Thomas Hofweber speaks of “Carnap’s rejection of ontology, and metaphysics more generally . . .”;4 Charles Landesman rou-tinely described Carnap as seeking to “debunk” ontology.5

This conception of Carnap’s goal—viz., his seeking to undermine ontology as tradi-tionally conceived—is widely shared. It is, I think, fundamentally mistaken. Carnap is engaged in a far more subtle effort: not to undermine ontology, but rather to portray it as consistent with empiricist epistemological scruples.

To cut to the chase: Carnap’s theory of ontology parallels noncognitivist theories of morality. Recall that such theories aim neither to eliminate nor to discredit moral evalu-ative practice, but rather to portray it as legitimate in light of possible metaphysical and/or epistemological misgivings. Carnap’s theory of ontology is best understood analo-gously: ontological discourse—discourse about what sorts of entities exist—is a device that enables explicit articulation of pragmatically motivated commitments to the adop-tion of certain linguistic forms. It is, in current parlance, an expressivist theory.6 Thus construed, the practice of ontology is fully consistent with empiricist requirements.

1 Le Poidevin (2009, 91). 2 Price (2011, 13).3 Eklund (2009, 130–56); see also his (2013, 229–49).4 Hofweber (2011).5 Landesman was one of my undergraduate metaphysics teachers. His lectures fostered the idea that

Carnap viewed traditional ontology as defective and unworthy of continued practice.6 Expressivism is one species of non-cognitivism. Helpful terminological clarifications, and contrasts

between expressivism and other varieties of non-cognitivism, are provided in Schroeder (2010: Chs. 2 and 4).

2Three Carnaps on Ontology

Robert Kraut


32 Robert Kraut

The theory that emerges from this understanding of Carnap generates extraordi-nary complexities—some of which find echo in other applications of expressivism, and others apparently unique to expressivism as applied to ontology. The goal here is to provide a systematic overview of Carnap’s much misunderstood theory of ontology, and to explore its prospects, deficiencies, and consequences.

The structure of the discussion is as follows: Section 2.1 motivates Carnap’s inquiry by articulating puzzles surrounding the notion(s) of existence as deployed in tradi-tional ontological theorizing. Section 2.2 distinguishes three likely interpretations of Carnap’s claims about ontology: eliminative, reductive, and expressive. It emerges that the third, expressivist interpretation provides the most plausible and philosophically rich—but also most daring—account of the discourse and practice of ontology. Subsequent sections fine-tune the theory, observe it in action, deflect likely objections, and tease out several undesirable consequences. Finally it is noted that Carnap’s account of ontological practice—construed as a non-cognitivist/expressivist theory—suffers from a peculiar circularity (here dubbed the “No Exit” problem) which is not obviously encountered in other such theories. Efforts are made to determine the extent—if any—to which such circularity undermines Carnap’s account.

2.1The existence of abstract entities—and the existence of moral facts, Fregean senses, possible worlds, Cartesian souls, fictional entities, numbers, qualia, arbitrary mereo-logical sums, colors, the future, or a host of other familiar targets of ontological dispute—remains controversial. Perhaps the disputed entities are problematic in their own right; but more baffling is the very content of the ontological controversy: what would it be for such items to exist, or to not exist?7

Not clear. Despite its familiarity, the basic concept of existence is puzzling: it is diffi-cult to say (without circularity) what the existence—or nonexistence—of various con-tested items amounts to.8 There are apparently non-problematic cases: disputes about the existence of caloric, or prime numbers greater than ten, induce little metaphysical

7 There might be no uniform answer: ‘exists’ might be equivocal and kind-specific (cp. Aristotle’s claim that ‘being’ is said in many ways, depending upon the category in which it is predicated). But compare David Lewis: “I do not have the slightest idea what a difference in manner of existing is supposed to be” (1986, 2). Regardless of the verdict concerning equivocity, the present inquiry remains viable: if ‘exists’ and cognate expressions are equivocal, the meanings of various kind-specific ontological notions still demand illumination.

8 Those who discuss ontology often bypass the root problem. Steven Yablo, for example, in discussing the ontology of sets, says

One side maintains with Putnam and Quine that indispensability of sets in science argues for their reality; the other side holds with Field and perhaps Lewis that sets are not indis-pensable and (so) can safely be denied. Either way, the point is to satisfy curiosity about what there is.

But it is difficult to say whether indispensability is a good criterion for the “reality” of sets unless one has an antecedent notion of what it is for a kind of entity to be real. See Yablo (1998, 229–61).


three carnaps on ontology 33

anxiety. But other cases—venerable controversies about abstract entities, colors, meanings, and the like—are less straightforward.

Perhaps the long-standing ontological controversies are illusory: different notions of existence might be at work, or no well-defined notion at all.

We might begin by attempting an explicit definition of ‘exist’:

There exist objects of kind K =df

K’s have explanatory utility (or: K’s are explanatorily indispensable); orK judgments are intersubjectively possible; orK’s are localizable in space-time; orK-facts are incorporable into a unified system; orK’s must be counted among the furniture of the world when taking inventory; orDiscourse about K’s is literal, rather than fictional;

And so on. The problem with all such reductive/definitional attempts is that they do not get at the essence of existence: they are either too broad, or too narrow, or circular. Moorean open question arguments make this clear: no conceptual confusions lurk in entertaining the possibility of existent entities that lack causal efficacy, explanatory indispensability, determinate spatio-temporal location, intersubjective accessibility, or position in a unified system. Moreover, reference to “furniture” and “inventories” provides little more than picturesque redescription; and the contrast between literal and fictional discourse seems to presuppose a distinction between real and fictional existence, thus putting us back where we started.

Analyses are not always possible; ‘existence’ and cognate expressions (such as ‘being’ and ‘reality’) might be so fundamental as to resist capture in any noncircular para-phrase. Perhaps existence is, as G.E. Moore said of goodness,

one of those innumerable objects of thought which are themselves incapable of definition, because they are the ultimate terms by reference to which whatever is capable of definition must be defined.9

If this is so, then perhaps the best we can do is make explicit the relevant inferential connections—for example, introduction and elimination rules for quantifier expressions—and leave it at that. We can shun definitional efforts, and simply treat ‘exists’ as primitive and non-problematic. (“Any sufficiently clear concept can be made primitive” [Dana Scott]).10 But the intractability of certain traditional meta-physical disputes—about the existence of abstract entities, for example—raises suspicion that more is required. Perhaps the concept of existence is not “sufficiently clear” after all.

If the goal is to better understand customary metaphysical discourse about exist-ence, several semantic approaches to first-order quantification theory might hold promise. Consider substitutional theories, according to which truth conditions for

9 Moore (1903, 61). 10 Scott (1970, 144).


34 Robert Kraut

quantified formulae are specified in terms of the truth of closed atomic sentences.11 On standard substitutional accounts, ‘(∃x)Fx’ is true if and only ‘Ft’ is true for some singu-lar term t contained within the substitution class. Perhaps existence is best understood in terms of truth. But perhaps not: for this substitutional truth condition presupposes a notion of existence (specifically, the existence of certain lexical items within the substi-tution class of singular terms) in order to get the substitutional recursions off the ground.

Or consider standard game-theoretic semantics, according to which ‘(∃x)Fx’ is true if and only if a winning strategy is available for the two person search-and-find game asso-ciated with the matrix ‘Fx’.12 Perhaps existence is best understood in terms of search pro-cedures and locatability. But perhaps not: for this game-theoretic truth condition presupposes a notion of existence (specifically, the existence of a winning strategy in the associated game) in order to get the game-theoretic recursions off the ground.13 Moreover the winnability of certain search-and-find games surely requires that the appropriate items exist (within the appropriate field of search) and are able to be found. Thus exist-ence is presupposed, rather than illuminated, by the game-theoretic approach.

Or consider straightforward objectual quantification. Perhaps the existence of Ks is best understood in terms of the truth of some sentence of the form ‘(∃x)Kx’, wherein bound variables carry ontological commitment. But here lurk familiar puzzles about fictional entities, Meinongian quantification, and other issues concerning relations between quantification and ontology. As Philip Bricker notes,

Non-existent entities can be quantified over, referred to, and truly attributed properties. Existence is not a prerequisite for being talked about. . . . Distinguishing a restricted quantifier that has existential import from an unrestricted quantifier that does not only has a point if existence is a substantial property that some things have and other things lack.14

The upshot—yet again—is that discussions of quantification and specification of truth conditions for quantified formulae assume, rather than illuminate, the concept of existence.

The general point is not that substitutional, game-theoretic or objectual semantics for quantifiers lack theoretical interest or applicability; the point is rather that such strategies presuppose the concept of existence and thus cannot provide the illuminat-ing, noncircular analyses we seek.

11 Dunn and Belnap (1968, 177–85); Marcus, “Quantification and Ontology” and “Nominalism and the Substitutional Quantifier,” both in her Modalities: Philosophical Essays (New York: Oxford University Press, 1993); Parsons (1971, 231–7).

12 See, for example, Hintikka, “Language-Games for Quantifiers” and “Quantifiers, Language-Games, and Transcendental Arguments,” both in his Logic, Language-Games, and Information (Oxford: Oxford University Press, 1973).

13 Neil Tennant denies this. His own implementation of game-theoretic semantics requires no quantifi-cation over strategies, thereby enabling him to claim that “The notion of game-winnability finds perfect (and inductively definable) expression without any explicit ontologizing, as would be involved if one were to equate winnability with there being a winning strategy.” See Tennant (2001, 3–20).

14 Bricker (2014).



There is another option: treat discourse about existence as irreducible and unanalyzable, but nevertheless illuminable as an instrument for expressing certain kinds of commitments, manifesting certain sentiments, or prescribing certain policies and/or plans of action. We might seek an account of the purposes typically served by talk of existence, or the conditions under which attributions of existence are assertible, or the circumstances that typically move standard speakers to attribute existence. Such an account takes the form of an expressivist explanation of ontological discourse—analogous to those metaethical theories that provide explanation rather than reductive analysis of moral discourse. Carnap had considerable sympathy for such explanatory strategies, despite the many puzzles they generate. The salient question becomes: What are we doing when we countenance the existence of a kind of entity?

2.2One way to understand the Carnap of The Logical Syntax of Language and “Empiricism, Semantics, and Ontology” is eliminative. Weary of centuries-old, incessant disputes about the objective existence of various kinds of entities, and propelled by his radical empiricism, Carnap offered a bold indictment:

An alleged statement of the reality of the system of entities is a pseudo-statement without cog-nitive content.15

If someone decides to accept the thing language, there is no objection against saying that he has accepted the world of things. But this must not be interpreted as if it meant his acceptance of a belief in the reality of the thing world; there is no such belief or assertion or assumption, because it is not a theoretical question.16

Such passages recommend—on one reading—the elimination of metaphysical dis-course and its replacement by a less problematic discursive device. Ontological dis-putes are deemed cognitively vacuous and a waste of time, and should thus be deleted from our repertoire. The motive behind such elimination is that the concept of “objec-tive existence” is not sufficiently well defined; therefore we should stop arguing about the existence of various kinds of entities and—as Carnap recommends elsewhere—argue instead about the practical consequences of adopting one or another “linguistic framework.” Carnap denies that there is such a fact as the objective existence (or non-existence) of a kind of entity—a fact that would legitimize adoption of one discursive framework rather than another. Ask not whether numbers enjoy objective reality: this is a pseudo-question. Ask instead about the practical consequences of adopting or abandoning first-order number theory. Such a “replacement” interpretation of Carnap’s strategy is embraced by William Demopoulos:

. . . Carnap’s goal in ESO is to show how the notion of a linguistic framework can be used to transform a traditional metaphysical problem into a problem of an altogether different

15 Carnap (1950, 214). 16 Carnap (1950, 207–8).


36 Robert Kraut

character. . . . Questions that are advanced as questions about reality, but that are not amenable to [resolution by ordinary methods] are without “cognitive significance” and should be trans-formed into questions about the choice of a language form.17

The recommendation that ontological discourse be replaced by (or “transformed into”) discourse about the pragmatic advisability of adopting specific linguistic frame-works is radical indeed. Most reflective theorists feel the urge to ask not only which linguistic forms ought to be adopted, given the practical and theoretical goals at hand, but also what there is. Indeed, the practical question cannot be answered without information about what there is: goals and strategies cannot be assessed in the absence of information about boundary conditions, and such information reintroduces the problematized concept of existence. The intuitive contrast between practical and onto-logical inquiries will not go away.

Put this another way: the proposed elimination/replacement borders on incoher-ence. For if questions can be raised about whether adoption of a given linguistic frame-work is warranted, the answer presumably requires, inter alia, reference to things that exist and the best way to deal with them. But the eliminative proposal disallows such justification, by disallowing talk about what exists. Apparently the “ontological elimi-nativist” has no access to the resources required for formulating reasons and justifica-tions for the envisaged revision in our discursive practices.

Thus the touted elimination does not seem feasible; besides, it is not clear why elim-ination of ontological discourse is advisable in the first place. Even if it be granted that assertions about existence cannot (without circularity) be translated into other terms, and even if there is no “sanitized” discursive idiom free of the problematized existen-tial concepts with which we might talk about existence and cognate concepts, nothing follows about the illegitimacy of those concepts. Other regions of discourse are simi-larly resistant to definitional-reductive analysis, without their disenfranchisement or elimination being thereby justified.

Put the soundness of the eliminative motive aside. This revisionary Carnap advo-cates radical departure from current practice, seeking to purge ontological inquiries from our repertoire. His eliminative mandate is: “Stop arguing about what kinds of entities exist; talk instead about what you can do for yourselves and how you can best do it.” But the price of such elimination is high: after the proposed discursive purge there remains an expressive impoverishment that frustrates the Ontologist in us all. For we wish to understand our world; this involves, among other things, knowing what there is. The proposed elimination disallows such inquiries.

Enter a second Carnap: more conservative, seeking to preserve ontological dis-course while proposing a conceptual analysis—a meaning-preserving paraphrase, or a reductive account of truth conditions—of such discourse. This Carnap claims that ontological questions really are pragmatic questions about the advisability of adopt-ing certain linguistic frameworks; they are “quasi-syntactic questions misleadingly

17 Demopoulos (2011, 647–69); here p. 653.



formulated in the material mode of speech.” Thus conceptual analysis, not elimina-tion, is the task at hand. This strategy, unlike the previous one, validates continued participation in ontological discourse, portraying it as less problematic than one might have feared. Here are some remarks that place Carnap within this camp:

To accept the thing world means nothing more than to accept a certain form of language . . .18

We may still speak . . . of “the acceptance of the new entities” since this form of speech is customary; but . . . this phrase does not mean for us anything more than acceptance of the new framework, i.e., of the new linguistic forms.19

Now it may be asked why I repeatedly propose to translate sentences which are formulated in the material mode of speech into the formal mode. I do this for the purpose of showing that such sentences belong to the field of syntax.20

On this view, ontological discourse is relatively harmless—when properly under-stood—and translatable, without loss of content, into discourse about acceptance of linguistic forms. Thus

(a) Natural numbers exist; they are items in the world.

is claimed to be meaning-equivalent (“translatable”) to

(b) Given our goals, it is pragmatically advisable to accept the framework of Peano Arithmetic.

The obvious problem is that the proposed reductive strategy is wildly implausible: it is doubtful that (a) and (b) are equivalent. Pending further argumentation, existence is one thing, pragmatically motivated acceptance of a linguistic framework quite another. Insofar as conceptual analysis seeks to provide plausible content-preserving transla-tions, prospects for analyzing talk of existence into talk of pragmatic advisability are grim.

Perhaps neither elimination nor conceptual analysis are well advised: discourse about the kinds of entities that exist might pack a unique expressive power, a power not to state facts or describe the world but to do something else. This suggests yet another strategy for illuminating ontological discourse.

Enter a third Carnap: a descendent of Hume, a close ally of the emotivists, advocat-ing a non-cognitivist or non-descriptivist account of ontological discourse: an explana-tion according to which it is not in the business of describing the world or stating facts.21 Ontological discourse serves some other purpose.

The suggestion is far-fetched: however the phrases ‘fact of the matter’ and ‘state of affairs’ are understood, surely it is a factual matter whether certain kinds of entities

18 Carnap (1950, 208). 19 Carnap (1950, 214). 20 Carnap (1935, 75–6).21 Here we embrace the orthodox assumption that non-cognitivism and non-descriptivism are two

sides of a single coin. This coupling is challenged by Terry Horgan and Mark Timmons, who allow for cognitive states with non-descriptive content. See Horgan and Timmons (2006, 255–98); their approach is soundly rejected in Schroeder (2008, 49–51).


38 Robert Kraut

exist: it is, for example, a fact that electrons exist, and that Newtonian absolute veloci-ties do not. Talk of existence appears to be paradigmatically descriptive: to say of what exists that it exists is to say what is true; to say of what exists that it does not exist is to say what is false.

Perhaps. But this Carnap, bent upon illuminating ontological discourse, is willing to flout appearances of descriptiveness: the role of such discourse, he claims, is to mani-fest, or render explicit, commitments to the adoption of certain linguistic forms.

This bold Carnap neither eliminates nor analyzes: he celebrates ontological dis-course as a useful instrument that serves to articulate commitments to certain lin-guistic resources. This Carnap does for metaphysics what emotivists do for morality, what Hume does for causation, and what Kripke’s Wittgenstein does for the language of rule-following: provide a non-reductive explanation that seeks to legitimize a region of discourse by portraying it as a non-descriptive mechanism for formulating commitments, expressing attitudes, or carrying out some other non-fact-stating task. Such explanation purports to be non-revisionary: conserving the discourse but legiti-mizing it (in light of concerns that motivate the inquiry) by providing a certain account of its role.

This “expressivist” Carnap is partly contrived; but there is historical basis. Noncognitivist explanations of moral discourse were part of Carnap’s intellectual cli-mate (in 1935 he notes that “a value statement is nothing else than a command in mis-leading grammatical form.”22) And the metaphysical discourse that constitutes the bulk of Heidegger’s Was ist Metaphysik? is construed by Carnap as expressive. Carnap says

The (pseudo)statements of metaphysics do not serve for the description of states of affairs. . . . They serve for the expression of the general attitude of a person toward life.23

Here we find Carnap’s expressivism made explicit. But even as an expressive mecha-nism, most metaphysics is deemed defective. It is deluded and self-deceived:

. . . through the form of its works it pretends to be something that it is not . . .The metaphysician believes that he travels in territory in which truth and falsehood are at

stake. In reality, however, he has not asserted anything, but only expressed something, like an artist.24

Thus metaphysics is more like art—a mechanism aimed toward expression—and less like science—a mechanism aimed toward articulation of truths about the world. But metaphysics is bad art: inferior to poetry, music, and other expressive art forms:

lyrical poets do the same without succumbing to self-delusion.25

The harmonious feeling or attitude, which the metaphysician tries to express in a monistic system, is more clearly expressed in the music of Mozart. . . . Metaphysicians are musicians without musical ability.26

22 Schroeder (2008, 25). 23 Carnap (1959, 60–81); here p. 78.24 Carnap (1959, 79). 25 Carnap (1959, 79). 26 Carnap (1959, 79).



Thus Carnap treats the bulk of metaphysics as a defective tool, an inadequate expres-sive substitute for art: it achieves—in misleading “theoretical” form—what art achieves more honestly and effectively.

But not all metaphysics is thus condemned: statements about the existence of sys-tems of entities are essential to semantic theory—a vital enterprise—and thus not eas-ily dismissed. Carnap’s “metalinguistic” pragmatism aims to legitimize ontology—at least, those portions of ontology required by semantic inquiries.

To sum up: this Carnap is a noncognitivist about ontological discourse. He advo-cates a “non-descriptivist” or “antifactualist” account of such discourse, analogous to noncognitivist accounts of moral discourse advocated by emotivists. This Carnap seeks neither elimination nor analysis of ontological claims: he wishes to explain them—while preserving their integrity—within the larger context of human commit-ment, as a mechanism that functions to achieve certain non-descriptive ends. Historical accuracy aside, this non-cognitivist Carnap deserves serious attention. Call him ‘Carnap*’.27, 28

2.3Here is Carnap*’s conception of ontology at work: providing a lens through which tra-ditional disputes might be viewed.

A) Consider a dispute about the existence of possible worlds. Two options are cus-tomarily available:

First Option: Possible worlds exist. They differ from the actual world in various ways, foremost among which is that we do not inhabit them.Second Option: The only possible world that exists is the actual world. Truth con-ditions for modal and counterfactual claims concern what goes on in this world (for this is the only world there is.) Granted, possible-worlds semantics is a helpful tool for exploring and modeling the structure of modal discourse; but the ontologically bizarre entities spawned by such semantic theories should be treated with instru-mentalist indifference.

Yet a third option—less frequently publicized—is bewilderment, grounded in uncer-tainty as to what it would be for possible worlds to exist (or to not exist).

27 Yet a fourth Carnap would be engaged in the task of explication: viz, “the task of making more exact a vague or not quite exact concept used in everyday life or in an earlier stage of scientific or logical develop-ment [here, the concept of existence], or rather of replacing it by a newly constructed, more exact con-cept . . .”. See Carnap (1947, 8–9). This fourth Carnap, though obviously related to his eliminativist counterpart, is not explored here.

28 Huw Price discusses a character, also dubbed (independently) ‘Carnap*’, who differs substantially from my own. My Carnap* is a traditional noncognitivist, whereas Price’s is not; mine seeks to conserve ontological practice, whereas Price’s does not; and, more generally, my Carnap* endorses the contrast between descriptive and expressive indicatives, whereas Price’s “global expressivism” rejects it. See Price (2011, 280–303).


40 Robert Kraut

Carnap* sees the ontological dispute as a clash of commitments concerning the prag-matic utility of Kripke-style semantics—broadly construed—in dealing with certain phenomena: specifically, in providing explanations of our customary ways of talking and thinking about possibility and necessity. The ontological claim that possible worlds exist is regarded by Carnap* as an expression of commitment to the pragmatic utility of adopting a specific linguistic framework: viz., that framework which accommodates talk of worlds, accessibility relations, similarity orderings, domains of individuals asso-ciated with worlds, mappings from worlds to extensions, denotation-at-a-world, truth-at-a-world, and other familiar notions of modal semantics. The skeptic about the existence of possible worlds is seen by Carnap* as denying the utility of such resources—believing them, perhaps, to be unwieldy and of dubious practical value—and thus as having undertaken a commitment to their eliminability. This latter commitment is explicitly represented in his/her insistence that possible worlds do not exist.

Such, at any rate, is Carnap*s configuration of what is going on. He makes no effort to minimize, ridicule, or debunk the dispute at hand: indeed, it is a real dispute—a clash of commitments—and an important one. But his diagnosis is conjoined with directives for a resolution: for his conception of ontological practice postulates under-lying contextual parameters relative to which each theorist’s metaphysical claims must be understood. If the theorists disagree about the data to be dealt with (modal dis-course? dispositional properties? nomic necessity?), or about criteria for successfully dealing with the data, or about criteria for the pragmatic advisability of adopting a specific linguistic framework, their apparent ontological disagreement about “the existence of worlds” emerges as no real disagreement after all. It is, rather, a manifesta-tion of their deploying different conversational contexts, with different underlying assumptions and goals. They are talking past one another; no wonder such a meta-physical “dispute” can go on interminably.

Thus Carnap* not only diagnoses the apparent intractability of the ongoing onto-logical dispute; he recommends a strategy for progress: encourage the participants to make explicit (1) the data they seek to deal with; (2) their sense of what it would be to adequately deal with it; (3) their criteria for treating one way of dealing with it as supe-rior to another.

“Dealing with the data” is a murky and evasive notion at best; for the present, think of these theorists as concerned primarily with the task of explanation. Thus construed, the ontological dispute about “the existence of possible worlds” is configured by Carnap* as a clash of commitments regarding the explanatory utility of the discursive resources of possible-worlds semantics.

Explanatory value and pragmatic utility are admittedly interesting; but Carnap* appears to have changed the subject: for these pragmatic considerations do not appear to capture the notion of existence that prompted our initial inquiry. An open question argument makes this clear: there could be possible worlds despite its not being prag-matically useful to speak of them; conversely, adoption of the linguistic framework of possible-worlds semantics might be pragmatically advisable despite there not being, as



a matter of metaphysical fact, any possible worlds. In a nutshell: there is a gap between considerations of what there is and what linguistic frameworks it is pragmatically advantageous to adopt.

Put this another way: Carnap*’s foray into context sensitivity, explanation, and pragmatic utility is orthogonal to the initial ontological puzzlement about whether there are, in fact, possible worlds. He has changed the subject.

It will emerge shortly, however, that Carnap* provides resources for acknowledging and sustaining this gap; despite initial appearances, he has not changed the subject.

B) Consider a dispute about the reality of mental events and psychological proper-ties. A Philosopher of Mind, impressed by considerations of explanatory/predictive power and systematic elegance, might wish to deploy a discursive framework that mobilizes the vocabulary and inferential resources of propositional attitude ascrip-tions. This commitment is expressed in her ontological claim that mental events and psychological properties exist. In contrast, her eliminativist opponent is committed to the adequacy of purely neurochemical explanatory resources, thereby leading him to deny any essential role to belief/desire attributions in adequate explanations of human behavior.

This ontological dispute about irreducibly mental events/properties is configured by Carnap* as a manifestation of conflicting commitments to the adoption of a specific discursive framework: viz., one that gives pride of place to causally efficacious and semantically evaluable internal states.

C) Consider a dispute about the existence of abstract objects: one faction alleging their existence, the other denying it. Carnap* portrays the customary confrontation between realist and nominalist as a clash of commitments to the explanatory utility of discursive resources that permit higher-order quantification and quantification over sentential posi-tions, thereby enabling semantic explanations that invoke propositions as referents of that-clause constructions and properties as referents of abstract noun phrases.

These examples, though vastly oversimplified, point toward Carnap*’s general strat-egy: portray conflicting ontological views as expressions of conflicting commitments to the adoption of specific linguistic forms. But an additional move has been made, for Carnap* offers more than a mere gesture toward expressivism. In the examples adum-brated here, he suggests a quite specific candidate for the kind of commitment explic-itly formulated in ontological claims: viz., commitments flowing from the perceived demands of explanatory projects.

Such an explanationist approach—in which the touted pragmatic considerations are grounded in the demands of explanation—contrasts with the more liberal approach of the historical Carnap, who explicitly demanded tolerance in permitting linguistic forms and the pragmatic criteria to be deployed in assessing their utility. Here, in con-trast, Carnap* privileges a specific pragmatic criterion—explanatory ineliminability—as playing the key role in ontological discourse.


42 Robert Kraut

Such privileging is curious: it does not appear that all pragmatic considerations are grounded in explanatory agendas. Explanation is only one endeavor likely to motivate the adoption or rejection of linguistic resources. It is, moreover, a bad idea to seek illu-mination of the concept of existence in terms of the pragmatics of explanatory efforts—if only because it is notoriously unclear what an explanation is.29 It is unclear, for example, whether a single notion of explanation applies across disciplinary bounda-ries, whether interest-relativity and idealization render explanation “subjective,” or whether explanation consists of derivation from covering laws, unification, or discov-ery of underlying causal mechanisms.

Carnap* acknowledges the host of puzzles surrounding the notion of explanation. Yet he insists—for reasons yet to be divulged—that disputes about the existence of a kind of entity are best construed as expressions of conflicting commitments to the explanatory ineliminability of a given discursive framework. And despite acknowl-edged complexities infecting the very idea of explanation, Carnap*’s strategy has obvi-ous merit. His rationale for privileging explanation-based considerations lies in their capacity to illuminate traditional ontological disputes. Construing the discourse of ontology as grounded in commitments to the explanatory ineliminability of a linguis-tic framework has the virtue of unifying a wide range of traditional ontological dis-putes: arguments about the existence of color manifest disagreements about the role of color predicates in psychophysical explanations; arguments about the existence of expressive properties of music turn on disputes about best explanation of music per-ception; arguments about the existence of a Judeo-Christian deity turn on disputes about best explanation of natural phenomena. And so on.

Whatever the virtues of Carnap*s theory as adumbrated thus far, it is vital to note—once again—that his portrayal of ontological disputes neither trivializes, debunks, or delivers quick resolution. Painful effort might be involved in determining whether a given discursive commitment is well advised on explanatory grounds—or, for that matter, on other sorts of pragmatically based grounds.

It is also vital to note that Carnap*’s theory deploys a contrast between adopting lin-guistic forms and adopting specific theories of the world; it thus assumes the familiar but maligned distinction between matters of language and matters of fact. His theory therefore conflicts with Quinean wisdom about “philosophy and science as [being] in the same boat . . .”30 For on Carnap*’s view there is a radical discontinuity between ontology and other theoretical frameworks: claims within physics and chemistry are about The Way The World Is, whereas claims within ontology are expressions of com-mitments to adopting linguistic forms. This contrast sits poorly with the ostensible similarity between ontology and other scientific/theoretical enterprises: the ontologist and scientist certainly appear to be involved in the same sort of endeavor—viz., describing the world. Moreover, the portrayal of ontological theories as implicitly

29 A fine historical survey is provided in Salmon (1989). 30 Quine (1969a, 127).



metalinguistic requires a version of the analytic–synthetic distinction, and is thus vul-nerable to familiar Quinean objections.31

All of this is true but unsurprising. Carnap* is, after all, a Carnapian. If countenanc-ing a contrast between metalinguistic considerations (about choice of linguistic frame-works) and factual considerations (about choice of scientific theories) packs the power to illuminate traditional ontological discourse, so much the better for the contrast.

But note in passing: it is not clear how strong a version of the analytic/synthetic dis-tinction is actually required by Carnap*’s strategy. Grant the difficulty in drawing any principled line (perhaps because there is none) between aspects of verbal behavior grounded in linguistic rules alone, and aspects grounded in collateral information: it is difficult to say when you mean different things by your words than I do, and when your meanings are the same but your background beliefs are outlandishly different. But Carnap* might have a theory of what we are doing when we chalk a dispute up to differ-ence of linguistic framework rather than difference of belief or background substan-tive theory. This is less a matter of requiring an analytic/synthetic distinction and more a matter of making sense of the way we explain our disagreements in terms of semantic differences rather than doxastic differences. So Carnap*, though feeling the force of Quinean criticisms, nonetheless has room to maneuver.

Despite complexities and potential difficulties, Carnap*’s theory illuminates the curious intractability of ontological disputes: refusal to affirm (or deny) the existence of a kind of entity—whether possible worlds, mental events, or arbitrary mereological sums—is seen by Carnap* as grounded in ambivalence about whether modal seman-tics, folk psychology, or unrestricted mereological composition are “genuinely explan-atory,” and thus whether commitment to the explanatory ineliminability of their resources is warranted.

2.4It is bad philosophical strategy to treat certain notions as so fundamental (or “basic”) as to resist analysis, while at the same time affording them such central significance as to require that their meaning be relatively clear to all. Carnap* is less than certain what existence consists of; but he refuses to be bullied into treating it as an “unanalyzable primitive.” Preferences for “desert landscapes” strike him not as ill-advised, but rather as unintelligible: for he does not know (for example) what it would be for abstract enti-ties to not exist, and thus how to distinguish metaphysically arid deserts from lush gardens. Thus his goal is to specify what people are doing when asserting or denying the existence of a kind of entity; his hypothesis is that ontological discourse is a mecha-nism that enables expression of commitments to the explanatory ineliminability of specific linguistic frameworks. Carnap* claims that his theory conserves customary ontological practice, in the sense that coming to believe the theory in no way undermines

31 See, for example, Quine (1966, 126–34).


44 Robert Kraut

continued engagement in the practice; it merely provides a better sense of what we are doing when thus engaged.

This is not an unfamiliar sort of claim, given non-cognitivist strategies elsewhere on the philosophical map. The moral expressivist—Carnap*’s counterpart in ethical theory—offers an explanation of moral discursive practice which purports to be con-sistent with continued, unrevised engagement in that practice: expressivism is touted as a conservative explanation that enables moralizing to go on as before.32 But it is not clear whether this is true: perhaps, having become convinced of the truth of expressiv-ism, an agent can never go back to moralizing in quite the same way as before. Perhaps the “higher level” explanation of the practice somehow trickles back down and under-mines the practice.33 Analogously, it is not clear that Carnap*’s strategy enables onto-logical practice to continue unaltered. Perhaps the self-understanding it provides would prompt metaphysicians to stop arguing about the existence of propositions—for example—and argue instead about the utility of adopting propositional quantifica-tion. But if this is a consequence of Carnap*’s theory then he fails by his own lights: for he sought a conservative explanation that facilitates comfortable acquiescence into ontological discourse, not its elimination or revision.

Apart from general misgivings about the conservativeness of non-cognitivist strate-gies, specific details of Carnap*’s theory merit skepticism. The theory privileges expla-nation as the foundation of ontological discourse; but obviously there is a contrast between existence and explanatory potency: natural numbers might exist even if arith-metic discourse is no essential part of any adequate explanation; Fregean senses might be real even if reference to them is not demanded by explanations of linguistic behav-ior or cognition; Platonic Forms might exist even if their ontological keep is earned via some route other than explanation; norms might exist even if they do not function as explainers. Some existents do not explain; and perhaps—depending upon the plausi-bility of “fictionalist” strategies—some explainers do not exist. There is more—and less—to the concept of existence than can be captured in terms of explanatory ineliminability.

Such observations, though compelling, bring us full circle: for they immediately prompt questions about what it would be for these items to exist, or to not exist—the very questions with which we began. At least Carnap*’s strategy provides foothold for an answer, whereas the alternative (viz., treating existence as “primitive”) is less than satisfying.

There is obviously work to be done in assessing Carnap*’s theory: the intuited con-trast between existence and explanatory ineliminability must be accommodated; the alleged conservativeness of Carnap*’s non-cognitivist theory—i.e., its allowing ontol-

32 See, for example, Blackburn (1993, 3–11).33 This “No Return” phenomenon—a global challenge to expressivist explanation—is systematically

explored in my “The Metaphysics of Artistic Expression: a Case Study in Projectivism,” in Johnson and Smith (2015, 85–105).



ogy to go on as before—must be addressed. And there are additional concerns, involv-ing various dimensions of ontological practice that Carnap*’s theory must address:

1) There are modal dimensions. The “realist” metaphysician, for example, sees the existence of abstract entities as transcendent, eternal and necessary: not dependent upon forces that condition pragmatically motivated commitments. But commitment and practical decision are permeated with contingency: projects might have been dif-ferent; conventions might have been different; explanatory goals might have been dif-ferent. We might not have found it advisable, given our goals and interests, to adopt (for example) the language of arithmetic. Carnap*’s suggestion that claims of existence are expressions of discursive commitments appears to get the modalities wrong: if such commitments might have been otherwise, then—it would seem—the existence of numbers might have been otherwise. But surely the existence of numbers, unlike the adoption of conventions and undertaking of commitments, is necessary. Carnap*’s theory thus fails to accommodate the modalities visible from within ontological practice.34

2) There are phenomenological dimensions. There is something it’s like to do ontol-ogy: one has the sense of engaging in discovery rather than invention. But if ontological claims are expressions of commitment, they are not announcements of discoveries; Carnap*’s explanation of ontology in terms of pragmatically motivated commitment fails to accommodate the phenomenology of ontological practice.

3) There are methodological and semantic dimensions. Expressivist theories—about morality, rule-following, or any other region of discourse—prompt ongoing skepticism. Peter Geach famously alleged “a radical flaw in this whole pattern of phi-losophizing,” insofar as it confuses predication and assertion and provides inadequate accounts of conditional embeddings; more recently, Mark Schroeder provides com-pelling arguments that “the prospects [for expressivism] are bleak.”35 Moreover, the purported bifurcation between “descriptive” and “expressive” sentences wreaks havoc upon compositional semantics: Boolean complexes of truth-conditional and expres-sive indicatives pose formidable challenges; sentences such as “If mass is conserved in all interactions, then spirits do not exist” and “If 11 is prime, then numbers exist” combine—according to Carnap*’s theory—descriptive antecedents with expressive consequents. Any holistically adequate theory of meaning must accommodate such hybrid constructions, but it is not clear how best to do so.

Many of these problems are general problems for expressivism: they are endemic to the entire non-cognitivist tradition with which Carnap* has cast his lot. Humean theo-ries of causation, emotivist theories of morality, deflationist theories of truth, Kripke’s Wittgenstein on rule-following, and related non-descriptivist strategies mobilize a contrast between fact-stating and non-fact-stating discourse; this contrast, in turn,

34 This objection was made especially clear to me in discussions with Christopher McMahon.35 Schroeder (2008: 15). See also Geach (1960: 221-5; 1965: 449-65) and Schroeder (2010).


46 Robert Kraut

requires a host of semantic, metaphysical, and psychological assumptions, none of which is self-evidently true. Perhaps Carnap* is injudicious in his choice of allies; nev-ertheless it is worth investigating whether his theory of ontological discourse suc-cumbs to challenges over and above those customarily encountered by expressivist theories: that is, whether his expressivist semantics for ontology encounters domain-specific problems resulting from special features of ontological practice.36 We consider the challenges in turn.37

2.5Carnap*’s theory apparently collapses the contrast between what there is and what lin-guistic forms it is advisable to adopt. But surely this contrast must be preserved by any adequate account of ontological practice.

Appearances notwithstanding, the contrast is not collapsed by Carnap*’s theory. For his theory does not portray ontological claims as conceptually equivalent to claims about the advisability of adopting certain linguistic resources (any more than expres-sivism portrays moral claims as conceptually equivalent to claims about sentiments). Carnap* preserves the contrast between what there is and what linguistic forms it is advisable to adopt, by exploiting the contrast between expressing a commitment and asserting the pragmatic advisability of undertaking that commitment.

To see this, an analogy is helpful. Consider the contrast between

(p1) I promise to meet you at Brenen’s tomorrow at noon.(p2) It is advisable, all things considered, that I meet you at Brenen’s tomorrow at noon.

These utterances (directed to my friend Lisa) are neither semantically nor pragmati-cally equivalent. (p1) performs the action of undertaking a commitment: incurring certain obligations, and licensing certain expectations on Lisa’s part; (p2), in contrast, does none of those things: it simply describes the situation as one in which I am better off meeting Lisa than not meeting her. This contrast precisely mirrors, according to Carnap*, the contrast between, e.g., claiming that propositions exist and claiming that it is advisable to adopt a language that permits propositional quantification: the former expresses a commitment, whereas the latter describes the world as making such a com-mitment advisable.

This contrast is familiar from discussions of performative linguistic acts. It is the actual undertaking of commitments, not judgments about the pragmatic utility of

36 The helpful contrast between domain-specific and domain-neutral problems confronting expressivism is deployed throughout Schroeder’s Noncognitivism in Ethics.

37 Yet another domain-specific challenge—not explored here—concerns ontological determinacy. According to Quine’s version of ontological relativity, there is no kind of entity such that adoption of a given linguistic framework mandates “acceptance” of that kind of entity: there is “slack” between accept-ance of a particular linguistic framework and commitment to the existence of a specific kind of entity. If Quine is right about this, then Carnap*’s theory falls short of explaining ontological practice; but Quine might not be right about this.



doing so, foregrounded in Carnap*’s theory. His conjecture is that there is, within our customary repertoire, an expressive mechanism that serves to render explicit certain discursive commitments we have actually adopted. That mechanism is the discourse of ontology.

The point is that Carnap* is no reductionist: claims about existence are not por-trayed as meaning-equivalent to claims about the pragmatic advisability of adopting linguistic forms. He preserves the semantic contrast between

(c) There are propositions.

and

(d) It is pragmatically advisable, in light of today’s projects, to adopt the language of propositional quantification.

(c) is not equivalent to (d): (d) is truth conditional, whereas (c) is not; (c) serves to explicitly formulate a commitment, whereas (d) claims that the facts are such as to warrant that commitment. Crudely: (d) describes the situation that legitimizes the commitment or decision embodied in (c).

Thus there is no collapse of talk of existence into talk of the pragmatic advisability of adopting certain linguistic forms. To this extent Carnap*’s theory is consistent with customary ontological practice.

2.6There are modal dimensions to the practice of ontology. If abstract entities exist, they exist necessarily; but necessity does not accrue to the commitments and practical deci-sions supposedly expressed by talk of existence: our commitments might have been different; but abstract entities could not have been different. Therefore existence claims cannot be construed as serving to manifest commitments to the explanatory inelimin-ability of linguistic forms: to think otherwise is to ignore metaphysically vital differ-ences of modal status. Carnap*’s theory fails to accommodate the modalities visible from within ontological practice.

That is the challenge. But the argument is unsound. The content of a commitment is distinct from the forces that prompt it. A commitment to studying number theory, for example, is a commitment to studying necessary relations, although the commitment itself is not necessary: one might have undertaken different commitments. Nothing in Carnap*’s theory mandates rejection of the idea that some objects and relations exist of necessity.

A related but more troubling challenge lurks in the vicinity: one that involves objectivity and mind-dependence of entities. An expressivist semantics for ontologi-cal discourse should not have the consequence that all items claimed to exist are mind-dependent. But if ontological claims are the expressions of attitudes, it is not clear how this consequence can be avoided.


48 Robert Kraut

It is helpful to consider an analogous objection frequently raised against expressivist theories of moral discourse. If the role of moral idioms is to manifest sentiments, then—the argument goes—morality itself is dependent upon sentiment. But then—the argument continues—so much the worse for expressivism, because the view from within moral practice countenances no dependence of morality upon sentiment. Immoral acts would be immoral even if everyone approved of them. Moreover: if the relevant moral attitudes are contingent upon various factors, then—the argument goes—moral properties are themselves contingent upon those factors; but this again flies in the face of prevalent intuitions about the objectivity of morality. Thus expressiv-ism fails to accommodate the phenomenology of customary practice, insofar as it—allegedly—portrays morality as mind- and sentiment-dependent.

Simon Blackburn suggests a strategy that enables the expressivist to circumvent such unwelcome results:

The utterance “whatever I or we or anyone else ever thought about it, there would still have been (causes, counterfactual truths, numbers, duties)” can be endorsed even if we accept the projective picture, and work in terms of an explanation of the sayings which gives them a sub-jective source.38

To see what Blackburn is up to, consider

(e) If everyone—myself included—had positive sentiments toward burning down orphanages, then burning down orphanages would be morally acceptable.

Expressivism does not validate (e): it does not underwrite the truth of conditionals that claim dependence of morality upon sentiment. Whatever the sentiments, burning down orphanages is morally unacceptable: in offering this latter verdict, the expressiv-ist sees himself—when turning self-reflective—as manifesting negative sentiments toward burning down orphanages, and also toward anyone’s having positive senti-ments toward such activities.

In other words: a carefully implemented expressivism purports to accommodate the possibility of an act’s being morally wrong independent of moral sentiment. The intu-ited gap between moral value and moral sentiment is explained as itself a manifesta-tion of moral sentiment.

Blackburn’s recommended procedure for avoiding the dependence of morality upon sentiment is tied to his more general semantic strategy of treating conditionals with normative antecedents and consequents as expressing “higher order attitudes”: attitudes about the co-tenability of attitudes. Unfortunately, his strategy has attracted considerable criticism (as Schroeder describes it, “[The] ‘higher-order attitudes’ accounts are plagued with fatal problems”39).

The analogy with expressivist accounts of moral commitment is instructive; fortu-nately, Carnap* requires none of the “higher order attitude” machinery deployed by

38 Blackburn (1984, 19). 39 Schroeder (2008, 9–10); fn. 3; see also van Roojen (1996, 311–35).



Blackburn, nor does his expressivism carry a commitment to Idealism: his theory does not entail the mind-dependence of existence. Consider

(f) If no one were committed to the explanatory ineliminability of discourse about microparticles, then microparticles would not exist.

If (f) were validated by Carnap*’s expressivist semantics for ontological discourse, the result would be the dependence of microparticles upon explanatory agendas. But Carnap*’s expressivist semantics does not underwrite the truth of (f); to see this, note that standard semantics for counterfactuals dictates the following intuitive procedure for assessing the truth of (f):

Go to the closest worlds in which people have different explanatory commitments than those commonly undertaken in the actual world—including commitments to the explanatory ineliminability of the linguistic framework of microphysics. Look at those worlds. You will see that in each of them, there are microparticles; never mind what the denizens of that world think about which linguistic frameworks are or are not explanatorily ineliminable. Each such world contains protons, neutrons, electrons, and all the other items we take to be constitutive of physical reality.

We might judge denizens of those worlds to be misguided in their linguistic commit-ments. But those non-mind-dependent entities that exist do not depend for their exist-ence upon anyone’s commitments. During reflective moments, I regard my ontological judgments about such situations as expressions of my own metalinguistic commit-ments. But that does not validate (f). What I would say were I a denizen of one of the worlds to which the antecedent of (f) directs attention is irrelevant. I assess these coun-terfactual situations “from the outside”.

To sum up: Carnap* denies that the semantic dependency of ontological claims upon explanatory commitments infects the content of such claims with contingency. His expressivism does not validate

(g) If everyone deemed it inadvisable to adopt the language of number theory, there would be no numbers.

Rather, his account validates

(g’) If everyone deemed it inadvisable to adopt the language of number theory, they would deny the existence of numbers.

Note that (g’) is consistent with standard ontological practice and with the necessary existence of numbers.

2.7There are phenomenological dimensions to the practice of ontology. One discovers in light of the evidence that Ks exist; only then does one regard the adoption of certain


50 Robert Kraut

linguistic forms as warranted. Theorizing about what exists involves the phenomenol-ogy of discovery, not that of decision; an adequate account of ontological practice must accommodate the “fact finding” phenomenology associated with it. Carnap*’s expres-sivist strategy, which explains ontological discourse as expressions of discursive com-mitment, seems ill-suited to explain the phenomenology of ontology. Here is a typical contrast between discovery and invention, and an underscoring of its importance, invoked in John Corcoran’s statement of “neutral platonism”:

. . . it holds that these objects [groups, sets, natural numbers, strings, etc.] exist and that they are independent of the human mind in the sense that

(1) their properties are fixed and not subject to alteration and

(2) they are not created by any act of will.

In a word: mathematical truth is discovered, not invented; mathematical objects are appre-hended, not created.40

Carnap* must explain why ontological inquiries appear to be in the business of track-ing down mind-independent and practice-independent realities, when (according to him) the most plausible account of those inquiries portrays them otherwise. This is a complex challenge; it is useful to place it on a larger map.

Consider Matthew, a set theorist interested in Banach–Tarski decompositions, ultrafilters, and large cardinal axioms. He wants to know what sets exist, of what size, and what properties obtain among them. He insists that when thinking about inacces-sible cardinals and other such exotica, he is involved in discovery, not invention. He denies that in doing mathematics he is merely grinding through the logical conse-quences of chosen axioms. He insists that his mathematical experience is that of exploring a realm of determinate set theoretic objects, much as an astronaut might explore the surface of the moon for determinate realities that lie hidden.

Matthew’s experience of mathematical reality constitutes phenomenological data that any adequate explanation of mathematical practice must accommodate. His broadly Platonistic picture—grounded, he says, in his own sense of “what proof and discovery are like”—is no irrefutable argument in favor of Platonism, any more than theistic reli-gious experience constitutes an irrefutable argument in favor of theism. But explanation is required of why Matthew sees the situation as he does. The point is commonplace: even if Platonism is false, we need to know why it seems true to so many working mathemati-cians. Matthew’s mathematical experience is a datum that must be accommodated—in one way or another—by any adequate philosophy of mathematics.41

Analogous phenomenological requirements operate in other philosophical enter-prises. Consider value: there is something it is like to discern good and evil in the

40 Corcoran (1973, 24).41 “The foundations of mathematics is, at least partly, a scientific study of mathematical practice. So

what mathematicians actually do and actually say is of direct interest to the foundations of mathematics.” See Harvey Friedman, Foundations of Mathematics weblog at <http://www.cs.nyu.edu/pipermail/fom/ 2006-April/010309.html>. See also my (2001, 154–83).



world. One has a sense of discovering—not creating or projecting—the rightness and wrongness of actions and states of affairs. The apparent objectivity of value is not, in itself, an obstacle to expressivist explanations of evaluative discourse; it is simply one more datum to be explained.

Indeed, the phenomenological dimension of participation in any practice must be accommodated by any purported explanation of that practice—expressivist or other-wise. And so it is with ontological discourse: ontological pronouncements do not “feel” like expressions of commitment; they feel, rather, like reports of discoveries. The above examples—mathematics, value, and modality—serve as reminders that the phenome-nology of participation in a practice is part of the data to be explained.

On this score Carnap*’s theory is no more problematic than any other non-cogni-tivist theory; indeed, he is well positioned for a reply. For he notes that a theorist dis-covers (rather than stipulates) the utilities associated with a given discursive framework, and discovers the consequences of any commitments she might under-take. One discovers what is (or is not) pragmatically advisable; one discovers whether a commitment brings other commitments in its wake; to this extent Carnap* is able to accommodate the “fact finding” phenomenology of ontology.42

There is, however, a lurking circularity infecting this strategy. Discovery that a given linguistic framework is pragmatically beneficial requires discovery that the framework facilitates transactions with the world and brings one closer to one’s goals. It is impossi-ble to make pragmatic assessments of a tool’s utility without reference to aspects of the world which the tool is intended to manage. Those aspects of the world exist: they pro-vide constraints upon efforts to meet one’s goals. Thus the notion of existence is deployed in the very process of assessing the pragmatic advisability of adopting a lin-guistic framework. Precisely this apparent circularity emerged earlier, in connection with the first Carnap’s eliminative strategy: elimination of a discursive framework—e.g., the framework of ontology—can be justified only in light of the way the world is, and arguments for eliminating a linguistic framework—e.g. the discourse of demonic pos-session—turn on considerations about things that exist in the world and relations that obtain among them. Precisely the same circularity apparently arises in connection with Carnap*’s expressivist strategy: the concept of existence is deployed in the very act of seeking to illuminate talk of existence in terms of pragmatically advisable adoptions of linguistic frameworks. There is, apparently, “No Exit” from talk of existence, even when the task at hand is to provide an explanation of the role played by talk of existence.

This problem arises in connection with other non-cognitivist strategies. John McDowell, for example, notes that non-cognitivist explanations of normative dis-course are likely to fail through circularity, given that any adequate explanation of such discourse must invoke not only causal relations but relations of appropriateness and warrant: thus an adequate explanation of normative discourse must, according to McDowell, avail itself of normative discourse, thereby leading to circularity.43

42 Here I have been helped by conversations with Kevin Scharp. 43 McDowell (1998, 131–50).


52 Robert Kraut

The problem is not intractable. Grant that pragmatic assessment of a linguistic framework assumes an existing reality for that framework to “‘push up against”.44 Carnap* accepts this: he claims a robust sense of mind-independent reality and vigor-ously denies any sympathy for idealism. He insists that the structure of an existing reality plays a key role in determining matters of pragmatic advisability. He thus deploys ontological discourse in his very theorizing about ontological discourse: it is, he says, because of what exists, combined with one’s goals and agendas, that adoption of one framework rather than another is pragmatically advisable.

But it is not obvious to Carnap* that deployment of the concept of existence when providing pragmatic assessments thereby vitiates his expressivist efforts. When he reflects upon earlier pragmatic decisions concerning adoption of linguistic frame-works, he explains his talk of existent objects and their impact upon pragmatic consid-erations in expressivist terms. He acknowledges that existent entities constrain his pragmatic deliberations—just as Hume acknowledges that lightening causes thunder, and just as emotivists acknowledge that wanton infliction of pain is morally reprehen-sible. After all, Carnap*’s expressivism was touted to be a conservative strategy, intended in part to earn us the right to continue doing ontology. If he could not avail himself of talk of existent entities when discussing pragmatic utilities, he would have failed by his own lights.

Put this another way: Carnap* claims to be a realist despite his expressivism about ontological discourse. The global challenge is to reconcile such realism with his prag-matist bent. This is explored at greater length in Section 2.9 below.45

2.8There are methodological and semantic dimensions to the practice of ontology. Ontological claims are coherently combinable with other sorts of claims. Descriptive claims frequently imply existential claims: “Joey Cat is on the desk, and desks are phys-ical objects” entails “Physical objects exist;” “7 is a prime number” entails “Numbers exist.” It is not clear how such entailments are possible, given that the first claim in each pair is descriptive and the latter expressive. Entailment relations require, at the very least, truth-evaluability of antecedents and consequents; insofar as ontological dis-course functions expressively, claims about existence are without truth value and thus cannot enter into logical entailment relations. Even if the theory is complicated—as it must be—to accommodate implication-like relations among expressive claims, the prospect of deriving expressive conclusions from descriptive premises presents a chal-lenge. Here is how Thomas Hofweber formulates the problem:

44 Thanks to an anonymous reader for suggesting this phrase and specific objection to Carnap*’s expressivism.

45 There is an unfortunate tendency to regard various forms of pragmatism as inconsistent with a robust realism. This is based upon confusions: pragmatism is not idealism, nor does it entail it. See Kraut and Scharp (2015, 331–60).



Existence claims are statements like “Numbers exist” or “There are numbers”. However, these statements are implied by ordinary mathematical statements like “There is a prime number between 8 and 12” or “Prime numbers exist between 8 and 123”. But how could an apparently purely factual mathematical statement imply an expressive one? It looks like either mathemat-ics itself is expressive, or there is a special notion of implication at work here. The former case generalizes, and ontological expressivism will thus imply global expressivism. The latter case is unacceptable, too. Implication doesn’t go that way. A factual statement does not imply an expressive one. This is completely contrary to what expressivism is all about . . .46

Recall that Carnap*’s theory is intended as an idealized and improved version of the theory suggested by Carnap. Carnap alleges an ambiguity in ‘Fs exist’: there is an “internal” and an “external” reading. On the internal reading, ‘Fs exist’ is entailed by ordinary discourse about Fs; but on the external reading it is not thus entailed: it is the latter, external reading that prompts Carnap to explore pragmatics and the adoption of linguistic frameworks, and prompts Carnap* to explore expressivism and commit-ments to explanatory ineliminability of discursive forms. Carnap*’s irrealism about ontological discourse is intended as a theoretically coherent implementation of Carnap’s views about external questions, and Carnap clearly regards ‘Fs exist’ as ambiguous.

Positing ambiguities is a bold strategy; it requires a principled account of the con-textual parameters that determine whether a particular tokening of an expression falls under one semantic interpretive rule or another. Carnap* might have no such account; the alleged ambiguity might prove implausible. But Carnap was optimistic about the contrast between internal and external existence claims, and such optimism is reflected in Carnap*’s non-cognitivist theory.

“Numbers exist” is implied by ordinary mathematical claims, and is illuminated by one’s chosen semantics for first-order quantification; but this leaves unclear, according to Carnap, the content of ontological claims made by realists and nominalists. The proper response to Hofweber is to note that the entailment relations to which he calls attention do not involve the ontological discourse treated expressively by Carnap*; whether such discourse can be effectively distinguished from its syntactically indis-cernible “internal” counterparts is yet another formidable challenge which Carnap* must confront.

2.9Finally, a serious methodological qualm—broached earlier—about Carnap*’s theory of ontology: it appears to deploy the very notion of existence it seeks to illuminate, thereby rendering it circular. It is difficult to avoid talk of existence, even when theoriz-ing about the role played by talk of existence. Call this the No Exit problem.

46 Thomas Hofweber, personal communication; but note his contribution to the present volume, wherein he denies that there are “two readings” of ‘Fs exist’.


54 Robert Kraut

Expressivist explanations—like any explanations—require ingredients for their implementation. A thoroughgoing expressivism about moral discourse, for example, requires a well-defined set of sentiments and/or stances—call it the projective base—elements of which are (according to the theory) expressed in moral indicatives. When Hume speaks of “gilding and staining natural objects with the colors borrowed from internal sentiment,” he assumes the existence of sentiments sufficiently rich to do the job, and “colors” which—whatever their origin—are real enough to get smeared onto the world. Emotivism exploits a set of “boo/hooray” attitudes; Kripke’s Wittgenstein, in explaining the role played by rule-following attributions, exploits feelings of confi-dence that an agent will continue a mathematical series in a certain way.47 In each such case, a certain phenomenological base is presupposed: the emotivist’s moral senti-ments, the Wittgensteinian’s feelings of confidence, etc. The expressivist, in suggesting that indicatives formulated with a given fragment of discourse serve to manifest stances, express commitments, or evince non-cognitive attitudes, needs a rich story about such stances, commitments, and attitudes, a story which does not backhandedly advert to the discourse under analysis.

But the expressivist also needs a story about the objects and events that serve as prompting stimuli to agents who manifest stances, project attitudes, and/or undertake commitments. Such objects and events must exist if they are to fulfill their explanatory role: after all, nonexistent objects cannot prompt speakers to undertake commitments. Thus the circularity: an adequate explanation of pragmatic decision and commitment requires deployment of ontological discourse; an expressivist explanation of ‘exists’ and cognate expressions requires engagement in discourse about what does and does not exist. Thus talk of existence is deployed in theorizing about talk of existence; we might say that there is “No Exit” from ontological discourse, even when the task at hand is the explanation of ontological discourse.

Whether this predicament vitiates Carnap*’s enterprise is not clear. On the one hand, if the goal is to theorize about certain commitments—those, for example, alleg-edly manifest in ontological discourse—there is reason to assume that such commit-ments have already been undertaken and thus permeate the theoretical enterprise: it is no surprise that talk of existence forces itself upon any effort to theorize about talk of existence. On the other hand, it is not clear that a puzzling concept can be illuminated by a theory that deploys that very concept. Carnap* might seek to soften the discom-fort by glibly invoking Neurath’s ship and reminders about Quine’s laudable (and fruit-ful) efforts “. . . to ponder our talk of physical phenomena as a physical phenomenon, and our scientific imaginings as activities within the world that we imagine.”48 But Quine’s scientific inquiry into science provides no helpful analogy: for Quine’s inquiry stems from no puzzlement about physical phenomena and scientific imaginings, whereas Carnap*’s inquiry stems from dismissive puzzlement about ontological dis-course. Acquiescing in a discourse while simultaneously questioning its legitimacy is

47 Kripke (1982). 48 Quine (1960, 5).



methodologically suspicious: if there is sufficient comfort with such discourse to deploy it in one’s theorizing, why the initial fuss?

But other considerations point in the opposing direction. Theories of language are formulated in language; inquiries into the semantics of questions are formulated as questions. Such considerations provide no warrant for treating the concepts of lan-guage or question as primitive, “unanalyzed” notions in all contexts. The fact that cer-tain investigations mobilize the very concepts under investigation is no argument for treating those concepts uncritically.

It is simply unclear how to theorize about aspects of our conceptual machinery when the very concepts to be illuminated are deployed in the theorizing. This predica-ment is familiar in other philosophical areas: validation of one’s norms of reasoning is problematic because those very norms are assumed in one’s efforts to validate them; reliability of one’s epistemic procedures is difficult to establish in any noncircular way because those very procedures are deployed in the process of establishing their relia-bility; evaluation of one’s theistic perspective is problematic because that very perspec-tive permeates one’s criteria of epistemic evaluation. Such situations offer No Exit from the concepts in question. Perhaps the fact—if it is a fact—that there is No Exit from ontological discourse provides a barrier to certain kinds of theories about such dis-course; or perhaps it will emerge, under more careful scrutiny, that the ontological concept problematized by Carnap* is not identical to the concept deployed in his back-ground theorizing, thereby dissipating the specter of circularity. But however these general methodological issues are resolved, they provide no basis for the presumption, deplored by Carnap*, that the concept of existence is sufficiently clear to warrant its deployment as a primitive, unproblematic notion.49

2.10Kit Fine notes that “there is a primitive metaphysical concept of reality, one that cannot be understood in fundamentally different terms.”50 In similar spirit, logician and com-puter scientist Arnon Avron recognizes “. . . basic concepts which cannot really be defined, and can only be explained in terms of themselves (or some equivalent

49 An illuminating alternative strategy for pursuing a broadly “naturalistic” metaphysics—perhaps bypassing the No Exit puzzle entirely?—is explored by Jenann Ismael. Her “side-on” strategy—inspired by Huw Price’s methodology—outlines a metaphysical method orthogonal to Carnap*’s. But upon close inspection it emerges that the No Exit problem remains:

In some cases, the side-on view will go like this: there are these things in the world and people come by and gather information about those things, and they form beliefs that are intended to reflect the way things are with the world . . .

This side-on metaphysical strategy thus assumes the very ontological notion (viz., there are these things in the world) that Carnap sought to illuminate; therefore the Price/Ismael method of metaphysics appears not to scratch precisely where Carnap itches. See Price (2011) and Ismael (2014).

50 Fine (2011).


56 Robert Kraut

notions),” noting that “There is no way to explain the quantifiers ‘for all’ and ‘exists’ without using at least one of these quantifiers . . .”51 Such sentiments bewilder Carnap*, thereby prompting his search for better understanding of ongoing arguments about the existence of various kinds of entities. Having despaired of meaning-preserving paraphrases or translational equivalents of ‘exists,’ ‘real,’ and cognate expressions, he seeks to clarify what we are doing when engaged in ontological discourse. His sugges-tion is that ontological claims manifest commitments to the explanatory ineliminabil-ity of specific linguistic frameworks. Despite the pitfalls of his theory—issues of circularity, conservativeness, and overall plausibility of expressivist strategies loom large—credit is due: Carnap*’s maverick willingness to confront what is perhaps the most fundamental question in metaphysics—the nature of existence—prompts a deeper look at the content of ontological disputes. If Carnap* is correct, a wary observer invited to participate in such disputes—whether about possible worlds, men-tal entities, universals, Fregean senses, or other popular targets of ontological contro-versy—should demand information about the data to be explained and the mode of explanation relevant to success. Should such demands be deemed irrelevant, the appropriate reaction to the ontological dispute is to walk away. Carnap’s legacy is to have earned us the right to do so.52

ReferencesAvron, Arnon (2007). “Concerning Definition of Formulas,” posted on FOM at <http://www.

cs.nyu.edu/pipermail/fom/2007-October/011993.html>.Blackburn, Simon (1984). Spreading the Word. Oxford: Clarendon Press.Blackburn, Simon (1993). Essays in Quasi-Realism. New York: Oxford University Press.Bricker, Phillip (2014). “Ontological Commitment”, The Stanford Encyclopedia of Philosophy,

Edward N. Zalta (ed.), <http://plato.stanford.edu/archives/win2014/entries/ontological- commitment/>.

Carnap, Rudolf (1935). Philosophy and Logical Syntax. Bristol, England: Thoemmes Press.Carnap, Rudolf (1947). Meaning and Necessity. Chicago: University of Chicago Press, 2nd ed.

1956.

51 Arnon Avron, “Concerning Definition of Formulas,” posted on FOM at <http://www.cs.nyu.edu/ pipermail/fom/2007-October/011993.html>.

52 Portions of earlier versions of this paper were presented in my seminar on Objects and Objectivity at Stanford University (Spring, 1996), Central Division Meetings, American Philosophical Association (Chicago, 2002), and an invited symposium on Carnap and Ontology (American Philosophical Association, Minneapolis, 2011). Thanks to Anthony Everett and Thomas Hofweber for criticisms and suggestions, and to Stewart Shapiro for tough but sympathetic exploration of the theory and its possible instabilities. Justin D’Arms provided helpful insights into relevant aspects of moral realism and “direction of explanation” problems; Christopher McMahon and Richard Creath helped me formulate some key points. Thanks also to Lisa Shabel, Julian Cole, Kevin Scharp, Matthew Foreman, and Jim Landesman for valuable comments on later versions. Special thanks to an anonymous reader for suggestions that led to improvements, and to Neil Tennant for exhilarating dismissals of Carnap*’s theory as incoherent.



Carnap, Rudolf (1950). “Empiricism, Semantics, and Ontology,” Revue Internationale de Philosophie 4: 20–40. Reprinted in the Supplement to Meaning and Necessity: A Study in Semantics and Modal Logic, enlarged edition (University of Chicago Press, 1956).

Carnap, Rudolf (1959). “The Elimination of Metaphysics Through Logical Analysis of Language,” in A.J. Ayer (ed.), Logical Positivism. Glencoe, IL: The Free Press.

Corcoran, John (1973). “Gaps Between Logical Theory and Mathematical Practice,” in Mario Bunge (ed.), The Methodological Unity of Science. Dordrecht: D. Reidel.

Demopoulos, William (2011). “On Extending ‘Empiricism, Semantics, and Ontology’ to the Realism/Instrumentalism Controversy,” Journal of Philosophy CVIII.

Dunn, J. Michael and Nuel D. Belnap (1968). “The Substitution Interpretation of the Quantifiers,” Nous II.

Eklund, Matti (2009). “Carnap and Ontological Pluralism,” in D. Chalmers, D. Manley, and R. Wasserman (eds.), Metametaphysics. New York: Oxford University Press.

Eklund, Matti (2013). “Carnap’s Metaontology,” Nous 47: 2.Fine, Kit (2001). “The Question of Realism,” Philosopher’s Imprint 1 (1) <www.philosophersim-

print.org/001001/>.Friedman, Harvey (2006). Foundations of Mathematics weblog at <http://www.cs.nyu.edu/

pipermail/fom/2006-April/010309.html>.Geach, P.T. (1960). “Ascriptivism,” Philosophical Review 69.Geach, P.T. (1965). “Assertion,” Philosophical Review 74.Hintikka, Jaakko (1973). Logic, Language-Games, and Information. Oxford: Oxford University

Press.Hofweber, Thomas. (2011). “Logic and Ontology,” in Edward N. Zalta (ed.), The Stanford

Encyclopedia of Philosophy, <http://plato.stanford.edu/entries/logic-ontology/>.Horgan, Terry, and Mark Timmons (2006). “Cognitivist Expressivism,” in Horgan and

Timmons (eds.), Metaethics after Moore. Oxford: Oxford University Press.Ismael, Jenann (2014). “Naturalism on the Sydney Plan,” in Matthew C. Haug (ed.), Philosophical

Methodology: The Armchair or the Laboratory? London and New York: Routledge.Kraut, Robert (2001). “Metaphysical Explanation and the Philosophy of Mathematics:

Reflections on Jerrold Katz’s Realistic Rationalism,” Philosophia Mathematica 9.Kraut, Robert (2015). “The Metaphysics of Artistic Expression: a Case Study in Projectivism,”

in R. Johnson and M. Smith (eds.), Passions and Projections: Themes from the Philosophy of Simon Blackburn. Oxford: Oxford University Press.

Kraut, Robert and Kevin Scharp (2015). “Pragmatism without Idealism,” in Christopher Daly (ed.), The Palgrave Handbook of Philosophical Methods. London: Palgrave Macmillan.

Kripke, Saul (1982). Wittgenstein on Rules and Private Language. Cambridge: Harvard University Press.

Le Poidevin, Robin (2009). “Existence,” in J. Shand (ed.), Central Issues of Philosophy. West Sussex, UK: Wiley-Blackwell.

Lewis, David (1986). On the Plurality of Worlds. Oxford: Basil Blackwell.Marcus, Ruth Barcan (1993). “Nominalism and the Substitutional Quantifier,” in Modalities:

Philosophical Essays. New York: Oxford University Press.Marcus, Ruth Barcan (1993). “Quantification and Ontology,” in Modalities: Philosophical

Essays. New York: Oxford University Press.


58 Robert Kraut

McDowell, John (1998). “Values and Secondary Qualities,” reprinted in Mind, Value, and Reality. Cambridge, MA: Harvard University Press.

Moore, G.E. (1903). Principia Ethica. Cambridge: Cambridge University Press.Parsons, Charles (1971). “A Plea for Substitutional Quantification,” Journal of Philosophy 68.Price, Huw (2011). Naturalism Without Mirrors. Oxford: Oxford University Press.Quine, W.V. (1960). Word and Object. Cambridge, MA: The MIT Press.Quine, W.V. (1966). “On Carnap’s Views On Ontology,” in The Ways of Paradox and Other

Essays. New York: Random House.Quine, W.V. (1969a). “Natural Kinds,” in Ontological Relativity and Other Essays. New York:

Columbia University Press.Quine, W.V. (1969b). “Ontological Relativity,” in Ontological Relativity and Other Essays. New

York: Columbia University Press.Salmon, Wesley C. (1989). Four Decades of Scientific Explanation. Minneapolis, MN: University

of Minnesota Press.Schroeder, Mark (2008). Being For: Evaluating the Semantic Program of Expressivism. Oxford:

Oxford University Press.Schroeder, Mark (2010). Noncognitivism in Ethics. London: Routledge.Scott, Dana (1970). “Advice on Modal Logic,” in K. Lambert (ed.), Philosophical Problems in

Logic. Dordrecht: D. Reidel.Searle, John (1987). “Indeterminacy, Empiricism, and the First Person,” Journal of Philosophy

LXXXIV.Tennant, Neil (2001). “Game Theory and Convention T,” Nordic Journal of Philosophical Logic

6: 3–20.van Roojen, Mark (1996). “Expressivism and Irrationality,” The Philosophical Review 105 (3).Yablo, Stephen (1998). “Does Ontology Rest on a Mistake?,” Proceedings of the Aristotelian

Society, supp. vol. LXXII.

Three Carnaps on Ontology - Northern Arizona University Three... · 7 There might be no ... rather...

Documents

Transcript of Three Carnaps on Ontology - Northern Arizona University Three... · 7 There might be no ... rather...